Twitter Safety @TwitterSafety
By removing viral Tweets from our data set (ten in total, none of which violated our policies), we see a very different picture: a consistent downward trend in true hateful language impressions. https://t.co/E9S3radJ6z — PolitiTweet.org